The Design Of Traffic Flow Forecasting System Based On Data Mining

In China, the rapid development of the highway has brought largely economic and social benefits as well as highway congestions, traffic accidents and environmental pollutions. There is a growing urgency to expect fast, safe, comfortable and efficient travel environment. Therefore, to improve service quality of highway management and other policy-making departments has not only become a trend, but also promoted a wider use of intelligent transportation systems.Intelligent Transportation System combines together three elements of the transport system-human, vehicle and road, so closely as to maximize transport and management efficiency of the whole transport system. Under intelligent transportation system, proper traffic flow forecasting is the basis of real-time traffic signal control, traffic assignment, route guidance, automatic navigation and the accident detection. To conclude previous traffic flow forecasting methods, they can be divided into methods based on traditional statistical theory, neural network, nonlinear theory and emerging technologies.In this paper, according to statistical theory, classification method in data mining technology and particular feature of traffic flow, proposes a variant classification method based on distance. The method holds that similar data values can be regarded as a "Class". In the prediction stage, the unknown value of a new case can be estimated by retrieving and computing the same "class" of records. In this paper, first, according to development goals of traffic flow forecasting system, makes a specific system design requirements analysis report. Secondly, under the guidance of data mining, make a detailed software design of traffic flow forecasting system. Finally, evaluate the actual performance of the traffic flow forecasting system.